50 Sir Edward Ffrencli Bromhead's Remarks on 



and also the tendency to cut off families at the limits of an 

 alliance for the sake of rounding the character. After the first 

 grouping the investigation of characters will, of course, be 

 found useful ; it will settle the limits of an alliance, where a 

 family has stood ambiguously between two; it will negatively 

 settle the place of many doubtful families; it will shake hete- 

 rogeneous alliances, and confirm such as are sound. 



The characters must be formed on the whole structure ; it 

 may in a few instances be possible to point out an assemblage 

 of families by some happy peculiarity, but such is not the plan 

 of nature, in which even species require an enumeration of 

 many parts. The utmost extent of simplification will be a 

 statement of the normal structure (the Nixus of Dr. Lind- 

 ley), followed by remarks on the limits of deviation and equi- 

 valent changes. The error in this case has arisen from the 

 received notion, that every species passes imperceptibly into 

 some other ; such, however, may not be the case, — a change of 

 one particular part generally implies, in natural history, some 

 corresponding change of every other part, so that families 

 may differ more abruptly than genera, and alliances more ab- 

 ruptly than families. In some families the differences of ad- 

 joining species are much greater than in others. 



A collection of characters is necessary also for ascertaining 

 the nature of the properties which usually extend over a suc- 

 cession of families ; the properties which most conveniently 

 distinguish genera are often of little value to distinguish fami- 

 lies, and those which separate adjoining families will probably 

 have little further range. In this research it is mortifving to 

 find a property running through the greater part of an alli- 

 ance, and often passed by in the remainder as being unimpor- 

 tant within the usual limits of classification. The characters 

 may too frequently amount to nothing, or indicate merely that 

 there is not a deviation from general normal structure ; but 

 such were some of the characters of Jussieu. They may 

 often be negative, but such are more definite than any other, 

 and may hereafter be pushed to any extent, so as to form dif- 

 ferential characters between any two alliances whatsoever. 



Results apparently trivial should be admitted, where they 

 offer themselves, as it is impossible to foresee what may serve 

 to illustrate the general progress of structure. Where a pro- 

 perty occurs through an alliance, with comparatively trifling 

 exception, it should be given, and the exception stated ; the 

 place of the excepted family may be shaken, or some remark- 

 able equivalence of structure may be discovered. 



Bartling very properly gives the lead to the structure of 

 the stem, foliation, and inflorescence, as Linnaeus does in the 



